Messages in this thread Patch in this message | | | Subject | [PATCH v2 4/4] serial/8250_pci: fix suspend/resume vs init/exit quirks | From | Dan Williams <> | Date | Mon, 09 Apr 2012 11:22:19 -0700 |
| |
Commit e86ff4a6 "serial/8250_pci: init-quirk msi support for kt serial controller" introduced a regression in suspend/resume by causing msi's to be enabled twice without an intervening disable.
00:16.3 Serial controller: Intel Corporation Patsburg KT Controller (rev 05) (prog-if 02 [16550]) Subsystem: Intel Corporation Device 7270 Flags: bus master, 66MHz, fast devsel, latency 0, IRQ 72 I/O ports at 4080 [size=8] Memory at d1c30000 (32-bit, non-prefetchable) [size=4K] Capabilities: [c8] Power Management version 3 Capabilities: [d0] MSI: Enable+ Count=1/1 Maskable- 64bit+ Kernel driver in use: serial
[ 365.250523] sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:16.3/msi_irqs' [ 365.250525] Modules linked in: nls_utf8 ipv6 uinput sg iTCO_wdt iTCO_vendor_support ioatdma dca i2c_i801 i2c_core wmi sd_mod ahci libahci isci libsas libata scsi_transport_sas [last unloaded: scsi_wait_scan] [ 365.250540] Pid: 9030, comm: kworker/u:1 Tainted: G W 3.3.0-isci-3.0.213+ #1 [ 365.250542] Call Trace: [ 365.250545] [<ffffffff8115e955>] ? sysfs_add_one+0x99/0xad [ 365.250548] [<ffffffff8102db8b>] warn_slowpath_common+0x85/0x9e [ 365.250551] [<ffffffff8102dc96>] warn_slowpath_fmt+0x6e/0x70 [ 365.250555] [<ffffffff8115e8fa>] ? sysfs_add_one+0x3e/0xad [ 365.250558] [<ffffffff8115e8b4>] ? sysfs_pathname+0x3c/0x44 [ 365.250561] [<ffffffff8115e8b4>] ? sysfs_pathname+0x3c/0x44 [ 365.250564] [<ffffffff8115e8b4>] ? sysfs_pathname+0x3c/0x44 [ 365.250567] [<ffffffff8115e8b4>] ? sysfs_pathname+0x3c/0x44 [ 365.250570] [<ffffffff8115e955>] sysfs_add_one+0x99/0xad [ 365.250573] [<ffffffff8115f031>] create_dir+0x72/0xa5 [ 365.250577] [<ffffffff8115f194>] sysfs_create_dir+0xa2/0xbe [ 365.250581] [<ffffffff81262463>] kobject_add_internal+0x126/0x1f8 [ 365.250585] [<ffffffff8126255b>] kset_register+0x26/0x3f [ 365.250588] [<ffffffff8126275a>] kset_create_and_add+0x62/0x7c [ 365.250592] [<ffffffff81293619>] populate_msi_sysfs+0x34/0x103 [ 365.250595] [<ffffffff81293e1c>] pci_enable_msi_block+0x1b3/0x216 [ 365.250599] [<ffffffff81303f7c>] try_enable_msi+0x13/0x17 [ 365.250603] [<ffffffff81303fb3>] pciserial_resume_ports+0x21/0x42 [ 365.250607] [<ffffffff81304041>] pciserial_resume_one+0x50/0x57 [ 365.250610] [<ffffffff81283e1a>] pci_legacy_resume+0x38/0x47 [ 365.250613] [<ffffffff81283e7d>] pci_pm_restore+0x54/0x87 [ 365.250616] [<ffffffff81283e29>] ? pci_legacy_resume+0x47/0x47 [ 365.250619] [<ffffffff8131e9e8>] dpm_run_callback+0x48/0x7b [ 365.250623] [<ffffffff8131f39a>] device_resume+0x342/0x394 [ 365.250626] [<ffffffff8131f5b7>] async_resume+0x21/0x49
That patch has since been reverted, but by inspection it seems that pciserial_suspend_ports() should be invoking .exit() quirks to release resources acquired during .init().
Acked-by: Alan Cox <alan@linux.intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> --- drivers/tty/serial/8250/8250_pci.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c index 024551a..24ea98c 100644 --- a/drivers/tty/serial/8250/8250_pci.c +++ b/drivers/tty/serial/8250/8250_pci.c @@ -2814,6 +2814,12 @@ void pciserial_suspend_ports(struct serial_private *priv) for (i = 0; i < priv->nr; i++) if (priv->line[i] >= 0) serial8250_suspend_port(priv->line[i]); + + /* + * Ensure that every init quirk is properly torn down + */ + if (priv->quirk->exit) + priv->quirk->exit(priv->dev); } EXPORT_SYMBOL_GPL(pciserial_suspend_ports);
| |